Control of the Morphology of Lipid Layers by Substrate Surface

Feb 24, 2014 - Control of lipid self-assembly on the surface was achieved by two different types of soft substrate materials, i.e., dextran and thiola...
2 downloads 15 Views 1MB Size
Download PDF
Article pubs.acs.org/Langmuir

Control of the Morphology of Lipid Layers by Substrate Surface Chemistry Niko Granqvist,*,† Marjo Yliperttula,† Salla Val̈ imak̈ i,‡ Petri Pulkkinen,‡ Heikki Tenhu,‡ and Tapani Viitala*,† †

Faculty of Pharmacy, Division of Biopharmaceutics and Pharmacokinetics, University of Helsinki, P.O. Box 56, 00014, Helsinki, Finland, and ‡Faculty of Chemistry, Laboratory of Polymer Chemistry, University of Helsinki, P.O. Box 55, 00014 Helsinki, Finland S Supporting Information *

ABSTRACT: In this study, surface coatings were used to control the morphology of the deposited lipid layers during vesicle spreading, i.e., to control if liposomes self-assemble on a surface into a supported lipid bilayer or a supported vesicular layer. The influence of the properties of the surface coating on formation of the deposited lipid layer was studied with quartz crystal microbalance and two-wavelength multiparametric surface plasmon resonance techniques. Control of lipid selfassembly on the surface was achieved by two different types of soft substrate materials, i.e., dextran and thiolated polyethylene glycol, functionalized with hydrophobic linkers for capturing the lipid layer. The low-molecular-weight dextran-based surface promoted formation of supported lipid bilayers, while the thiolated polyethylene glycol-based surface promoted supported vesicular layer formation. A silicon dioxide surface was used as a reference surface in both measurement techniques. In addition to promoting supported lipid bilayer formation of known lipid mixtures, the dextran surface also promoted supported lipid bilayer formation of vesicles containing the cell membrane extract of human hepatoblastoma cells. The new dextran-based surface was also capable of protecting the supported lipid bilayer against dehydration when exposed to a constant flow of air. The wellestablished quartz crystal microbalance technique was effective in determining the morphology of the formed lipid layer, while the two-wavelength surface plasmon resonance analysis enabled further complementary characterization of the adsorbed supported lipid bilayers and supported vesicular layers.

1. INTRODUCTION An understanding of different biochemical interactions present in human physiology is one of the primary goals of modern biochemistry and pharmaceutical research. Many of the methods and approaches for studying biochemical interactions use full systemic experiments on either human or animal physiology. Human or animal experiments are difficult to perform from both an experimental and an ethical point of view. In particular, analyzing in vivo results as well as evaluating different causalities is often very complicated. Therefore, both the study of biochemical interactions and current pharmaceutical development rely heavily on different in vitro experimentation.1,2 Most of these in vitro methods are either relatively simple and do not mimic the in vivo situation very well, such as the parallel artificial membrane permeation assay (PAMPA),2,3 or quite complicated, making control of the system difficult, as is the case with most cell-screening assays.4,5 Furthermore, these in vitro methods often use fluorescently labeled compounds, do not enable measurements in real time, and rely on secondary detection techniques. The overall efficiency for drug development is decreasing constantly (i.e., cost per new drug entering the market is increasing), partly because effective transfer of new formulations from laboratory scale to the clinic is very slow.6 Hence, there is a clear need for new, more efficient, and/or cost-effective methods for screening © 2014 American Chemical Society

biochemical interactions that could provide complementary information to already existing in vitro methodologies, as well as reduce the need of ethically questionable in vivo studies. Supported lipid membrane structures, i.e., supported lipid bilayers (SLBs) and supported vesicle layers (SVLs), are excellent biomimetic systems because they closely resemble cell membranes and other biological barriers consisting mostly of phospholipids. It is possible to incorporate membrane proteins, receptors, and other biologically relevant molecules into these lipid membrane structures to mimic biological membranes with specific functionalities.7 For example, it is possible to prepare asymmetrical SLBs so that the mobility of the lipids and other components is very near to that of the membranes found in the nature. The biological similarity of such cell membrane models can be excellent in the context of interaction research.8 Traditionally SLBs and SVLs are prepared directly on a solid support by vesicle fusion and vesicle adsorption, respectively.9 This method is often used in combination with several labelfree detection techniques, such as quartz crystal microbalance (QCM), dual polarization interferometry (DPI), and surface plasmon resonance (SPR). Received: December 4, 2013 Revised: February 21, 2014 Published: February 24, 2014 2799

dx.doi.org/10.1021/la4046622 | Langmuir 2014, 30, 2799−2809

Langmuir

Article

cholamidopropyl)dimethylammonio]-1-propanesulfonate) were obtained from Avanti Polar Lipids (Alabama, USA). Ethanol (95%) was obtained from Altia Corp. (Rajamäki, Finland). Hellmanex II was obtained from VWR Finland (Helsinki, Finland). All water used in the experiments had a resistivity of 18.2 MΩ/cm. The PEG-SAM was synthesized from a commercial PEG45-polymer with OH end groups and a molecular weight of 2 kD, obtained from Polysciences Inc. (Pennsylvania, USA). The synthesized PEG-SAM had the general structure of H2CCH(CH2)9PEG(CH2)10 SH or an analogous disulfide dimer. Reagents and methods used in the synthesis of the PEG-SAM are described in detail in the Supporting Information, section S1. Gold- and SiO2-coated SPR sensors were obtained from BioNavis (Ylöjärvi, Finland), and similarly coated 5 MHz QCM crystals were obtained from Q-Sense Inc./BiolinScientific (Västra Frö lunda, Sweden). 2.2. Surface Synthesis. Two different lipid-binding surface coatings were synthesized on gold-coated sensors for SPR and QCM measurements. The surface coatings on both the SPR and the QCM sensors were simultaneously prepared in the same reaction vessels in order to have as identical coatings as possible for both sensors. The dextran-based surface was selected because the polysaccharide structure is close to the natural polysaccharides found in the cell membranes. The small 6 kDa size should make relatively thin and dense layers upon grafting to a surface, and there are suitable chemical pathways to postmodify it into a lipid-anchoring surface.20,21 The PEG-SAM was selected because PEG is well known for its biocompatibility and low interaction with SLBs, and there are suitable chemical pathways which allow end-group modification of the PEG prior to grafting it to the support surface (see Supporting Information S1). Both of these polymers have been shown to work well with SLB and SVL layers in previous literature.8 The dextran-coated (Dex6 kDa) surface in this study was synthesized with slight modifications as described by Summanen et al.,20 which is a modification of the method by Löfås and Johnsson,21 i.e., 6 kDa Dextran at a concentration of 300 g/L was used instead of 500 kDa dextran at a concentration of 30 g/L, and 2 M bromoacetic acid was used instead of 0.5 M bromoacetic acid. The sensors were first cleaned by keeping them in a boiling H2O2:NH3:H2O (1:1:5) solution for 10 min and then washed thoroughly with ultrapure H2O. Hereafter, the sensors were immersed in a solution of 5 mM mercaptoundecanol in an 8:2 ethanol:water solution for 24 h. The sensors were then allowed to react for 3 h with epichlorohydrin (2% v/ v) in 0.1 M NaOH rinsed with water, transferred to a 300 g/L solution of dextran in 0.1 M NaOH, and left to react for 24 h. After this the sensors were washed thoroughly with ultrapure H2O and immersed in 2.0 M bromoacetic acid in 2 M NaOH for 24 h, after which the sensors were thoroughly washed with ultrapure H2O and stored at +4 °C. The surface-immobilized carboxylated dextran was then functionalized with decylamine by utilizing EDC/NHS activation chemistry. The carboxylated dextran surface was first activated by treating it for 10 min with a 80 mg/mL:20 mg/mL EDC/NHS in PBS buffer. Immediately after this, the activated dextran surface was treated with a 30 mg/mL decylamine suspension in PBS for 10 min. The surface was finally deactivated with a 1 M ethanolamine solution, washed thoroughly with ultrapure H2O and ethanol, dried, and stored dry in +4 °C for later use. The PEG-SAM surface coating was prepared on gold-coated sensors using a simple self-assembly protocol. The sensors were first cleaned by keeping them in a boiling H2O2:NH3:H2O (1:1:5) solution for 10 min, then washed thoroughly with ultrapure H2O, and dried with nitrogen gas. The sensors were then immersed in a 2 mg/mL ethanol solution of the PEG-SAM for 24 h. After formation of the selfassembled PEG-SAM layer the sensors were finally washed thoroughly with ethanol, dried carefully with nitrogen, and stored dry in +4 °C for later use. 2.3. Cell Membrane Extract Preparation. Cell membrane extract was produced from human hepatoblastoma cells (HepG2) derived from the liver tissue of a 15 year old male (HB-8065, ATCCLGC Promochem, USA). Cells were cultured in high-glucose

Common applications of cell membrane model systems are in membrane biophysics for SLBs9 and in studying biochemical interactions between soluble compounds and lipids or membrane proteins for SVLs.10 The SLBs and SVLs have also been introduced for pharmaceutical research.8,11,12 However, there are still several challenges such as control of the morphology between SLB and SVL, incorporation of membrane proteins and other membrane components of interest, as well as method robustness and lack of assays showing all these three properties together. Control of morphology has been extensively studied on inorganic supports, for example, SLBs are readily formed on SiO2 substrate and SVLs on Au and TiO2.9,13 Also, specific linking chemistry, such as biotin−avidin, HisTag, nucleotides, and ionic interactions between polyelectrolytes and charged lipids, has been used to bind lipid bilayers on solid support.14,15 However, often the challenge with both SLBs and SVLs is that the formed lipid layers are unstable and sensitive for irreversible denaturation of the membrane structure upon transition through the air−water interface.16 Improvements in the stability of SLBs against this denaturation have recently been demonstrated using cholesteryl-functionalized hydrated polymer supports,17 strongly interacting metal chelated SLBs,18 and by “sandwiching” SLBs between the support and proteins16 or PEG.19 In the case of using linking chemistry in formation of SLBs or SVLs one of the counterparts is also left on top of the final lipid layer,14 which can influence the surface properties and interactions of the lipid layer. The advantage of using hydrophobic linkers such as alkane chains or cholesterol immobilized on the support is that they do not introduce unwanted counterparts on top of the lipid layer.14,17 In this study we focused on improving the control of the morphology of the supported layer (SLB vs SVL) as well as the air stability of supported lipid membranes using polymeric supports. Furthermore, our aim was to enable an SLB formation of vesicles composed of known lipid mixtures or a mixture of cell membrane extract of human hepatoblastoma cells (HepG2) and a known lipid mixture in order to prepare SLBs with high biological relevance for biochemical and pharmaceutical applications. The support materials for anchoring either SLBs or SVLs were loose networks of hydrated polymers functionalized with hydrophobic linkers acting as a cushion between the supported lipid membrane structure and the substrate. The surfaces were prepared from commercially available dextran and a custom-synthesized thiolcontaining polyelthylene glycol−polymer (PEG-SAM). Silicon dioxide (SiO2) surface was used as a reference support due to its well-known ability to promote SLB formation by vesicle spreading.9 Three different known lipid mixtures and one cell membrane extract composition were used to study the type of lipid structure that was promoted by the different supports.

2. MATERIALS AND METHODS 2.1. Materials. Sodium chloride, sodium hydroxide, calcium chloride, 50% hydrogen peroxide, concentrated ammonia, HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid), 11-mercaptoundecanol, epichlorohydrin, bromoacetic acid, decylamine, EDC (1ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride), NHS (N-hydroxysuccinimide), ethanolamine, PBS (phosphate-buffered saline, tablet P4417), and Dextran (Mr = 6 kDa from Leuconostoc spp.) were obtained from Sigma-Aldrich (Helsinki, Finland). B-03labeled biotin was obtained from Episentec Ab (Sollentuna, Sweden). EggPC (egg phosphatidylcholine), POPS (palmitoyl-oleyl-phophatidylserine), cholesterol (from lamb wool), and CHAPS (3-[(32800

dx.doi.org/10.1021/la4046622 | Langmuir 2014, 30, 2799−2809

Langmuir

Article

Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum, 100 μg/mL penicillin, 100 μg/mL streptomycin, 2 mM L-glutamine, and 100 mM sodium pyruvate in a humidified atmosphere containing 5% CO2 at 37 °C. About 0.86 g of HepG2 cells was allowed to melt on an ice well. Cell pellets were suspended into a harvest buffer (50 mM Tris-HCl, 300 mM mannitol, pH7). The cell suspension was centrifuged at 800g for 5 min at 4 °C, and the supernatant was thrown away. This step was repeated twice. The residue was suspended in a membrane buffer (50 mM Tris-HCl, 50 mM mannitol, and 2 mM EGTA, pH7). Cells were then homogenized using a cell homogenizer by 40 strokes and incubated on ice for 1 h. The solution was centrifuged at 800g for 10 min at 4 °C, and the supernatant was taken and further centrifuged at 15 000g for 1 h at 4 °C. Again, the supernatant was taken and centrifuged at 100 000g for 75 min at 4 °C. Finally, the supernatant was removed, and the pellets were weighed and stored at −75 °C until used. A qualitative test for the presence of proteins in the extracted cell membranes was performed using a Bio-Rad Protein Assay Reagent kit (cat. no. 5000006), which verified that the cell membrane extract contained a substantial amount of protein. 2.4. Liposome Preparation. Three different lipid compositions were used for the liposomes for the SPR and QCM studies. EggPC was chosen as the majority of the lipids in prokaryotic cell membranes are phosphatidylcholines,22 and it offers a good combination of saturated and nonsaturated lipids that are physiologically relevant. The POPS was added to the lipid mixture to simulate the natural negative charge that is exhibited in natural prokaryotic cell membranes. Cholesterol was added to the lipid mixture in order to mimic the natural cell membranes even further as it is one of the major components controlling the rigidity of cell membranes. Liposomes were prepared with the sonication method.23 The lipid compositions were 100% EggPC, 75:25% EggPC:POPS, and 70:25:5% EggPC:POPS:cholesterol by molar ratio. A labeled liposome containing 100% EggPC (EggPC+Label) liposome and 1 μg/mL of B-03-labeled biotin in the hydration buffer was also prepared for SPR experiments. Shortly, the liposomes were prepared by first drying the lipid mixture with a nitrogen flow and rehydrated to a total lipid concentration of 1 mg/mL with a HBS + Ca2+ buffer (20 mM HEPES, 150 mM NaCl, 3 mM CaCl2, pH 7.4). Hydration was allowed to take place for 1 h with mixing every 15 min. After hydration, the liposomes were sonicated with a Vibra-Cell VCX 750 sonicator (Sonics & Materials Inc., Newtown, CT, USA) in an ice−water bath until the solution became clear, which typically took 5−15 min depending on the lipid mixure. The quality of the prepared liposome solution was always checked with a Malvern Zetasizer 3000HSA dynamic light scattering (DLS) instrument (Malvern Ltd., Malvern, UK). A hydrodynamic radius below 50 nm and a monomodal size distribution for the liposomes were required in order to accept the prepared batch for later use. The number-averaged size of the liposomes obtained from DLS measurements were 27 nm for EggPC, 27 nm for EggPC+PS, 39 nm for EggPC+PS+Chol, and 27 nm for the EggPC+Label. CONTIN analysis of the size distributions is presented in the Supporting Information (Figure S3). Liposome stock solutions were stored at +4 °C and used within 5 days. Labeled EggPC liposomes were prepared exactly as the other liposomes, except that 1 μL of 1 mg/mL B-03-labeled biotin in pure water solution was added to the HBS + Ca2+ hydration buffer during the hydration step. Lipsomes containing the HepG2 membrane extract were prepared with a method adapted from the work of Dodd et al.24 The HepG2 membrane extract from the centrifugation pellet was dispersed in a HBS + Ca2+ buffer to form a 10 mg/mL stock solution and then sonicated for 20 min in the same conditions as the other liposomes in this study. This extract was then mixed with a 10 mg/mL presonicated stock solution of EggPC liposomes with a mass ratio of 4:10 extract:EggPC forming a total lipid concentration of 0.14 mg/mL. This solution was then sonicated for 5 min at the same conditions as described before for other liposomes in this study in order to mix the different liposome populations. The sonicated membrane extract mixture was always used for the QCM and SPR measurements within

the same day of preparation. DLS measurements of the resulting HepG2-EggPC liposomes gave a number-averaged size of 63 nm. The HepG2 cell membrane extract was mixed with EggPC because in general natural total extracts do not fuse into SLBs due to the stability of natural extracts, but addition of more unstable lipid such as EggPC can results in a bilayer formation, as described by Dodd et al. 2.5. QCM Measurements. QCM measurements were performed with a KSV QCM-Z500 instrument (KSV Instruments, Helsinki Finland) at 20 °C. The flow rate used for measurements was 250 μL/ min, and the running buffer was HBS (20 mM HEPES, 150 mM NaCl, pH 7.4). The frequency and dissipation changes for the third, fifth, seventh, and ninth overtones (marked from here on as F3, F5, F7, and F9) were recorded during all experiments. The results were analyzed using the KSV QCM-Z500 software (version 3.4). Silicon dioxide QCM sensors were washed in situ in the flow channel with sequential 5 min injections of 20 mM CHAPS, 2% Hellmanex II, 95% ethanol, and ultrapure H2O.25 Experiments were performed by first measuring a baseline with the running buffer for approximately 10 min and then injecting a 0.1 mg/mL liposome solution for 8 min followed by a 10-min rinse period with the running buffer. Samples were measured consecutively with a wash sequence between the samples. The Dex6 kDa and PEG-SAM surfaces were washed in situ in the flow channel before measurements by injecting 20 mM CHAPS for 5 min followed by rinsing with ultrapure H2O. The measurement sequence was the same as that for the SiO2-coated QCM sensors, with the exception that a 5-min ultrapure H2O injection was added after the flush period with the running buffer, followed by a second flush period with the running buffer. The QCM thickness analysis of the lipid layers was performed using either the Sauerbrey or a viscoelastic model (equivalent circuit analysis) for the deposited lipid layers.26 It was necessary to use two different models for QCM analysis, because the Sauerbrey model describes the thickness of a rigid layer (such as an SLB), and the equivalent circuit analysis provides the thickness and mechanical properties of a viscoelastic layer. Constant parameters used in the modeling were as follows: density of lipid/liposome layer 1.0 g/mL, density of buffer 0.9986 g/mL, and viscosity of buffer 0.890 mPa s. 2.6. SPR Measurements. SPR measurements were simultaneously performed at wavelengths of 670 and 785 nm with an MP-SPR instrument SPR Navi 200-L (BioNavis, Ylöjärvi, Finland). Measurements were performed at 20 °C with a flow rate of 30 μL/min. The flow rate for the SPR measurements was selected so that the hydrodynamic flow conditions in the SPR flow channel matched the flow conditions in the QCM flow channel.27 Theoretically, the SPR flow channel used in this study would have a 9 times higher surface shear stress than the QCM flow channel if the same flow rate was used in both flow channels. SiO2 SPR sensors were washed in situ in the flow channel with sequential 3-min injections of 20 mM CHAPS, 2% Hellmanex II, 95% ethanol, and ultrapure H2O. Experiments were performed by first measuring the baseline with the running buffer for approximately 10 min, injecting a 0.1 mg/mL liposome solution for 8 min, followed by a 10-min rinse/flush period with the running buffer. Samples were measured consecutively with a wash sequence in between the samples. The Dex6 kDa and PEG-SAM surfaces were washed in situ in the flow channel before measurements by injecting 20 mM CHAPS for 5 min. SPR measurements did not need the same ultrapure H2O rinsing treatments as the QCM measurements between consecutive experiments. The air stability of the EggPC+PS+Chol lipid bilayers on Dex6 kDa was measured by running air for 10 min at a flow rate of 200 μL/min over the SPR surface. After exposing the lipid bilayer to air, the surface was rewetted with the same HBS buffer, and the return of the baseline was recorded. The full SPR angular range was continuously monitored during this period, and the change to air and back to the buffer was confirmed in real time from the critical angle (total internal reflection angle, TIR from now on) shift induced in the SPR angular spectrum when going from water-based buffer to air and back to the buffer. 2801

dx.doi.org/10.1021/la4046622 | Langmuir 2014, 30, 2799−2809

Langmuir

Article

Figure 1. QCM Δf/N vs time sensograms at F3, F5, F7, and F9 for three known lipid formulations deposited on SiO2 sensor. (Left to right) Red, EggPC; black, EggPC+PS; blue, EggPC+PS+Chol. Horizontal line marks the position where Δf/N = 25 Hz. Arrows indicate events in the sensogram: (a) injection of liposomes, (b) end of injection of liposomes. Final frequency change (25 Hz) and overlap of the overtones indicate that the liposomes rupture and form lipid bilayers on the SiO2 sensor surface.

Figure 2. QCM Δf/N vs time sensograms at F3, F5, F7, and F9 for all four vesicle formulations used in this study during interaction with the Dex6 kDa surface. (Left to right) Red, EggPC; black, EggPC+PS; blue, EggPC+PS+Chol; green, HepG2-extract. Horizontal line marks the position where Δf/N = 25 Hz. Arrows indicate events in the sensogram: (a) injection of liposomes, (b) end of injection of liposomes, (c) injection of water, and (d) end of injection of water. Bilayer formation was triggered by treating the deposited lipid layers with H2O. Final frequency change (25 Hz) and overlap of the overtones indicate that the liposomes rupture and form lipid bilayers on the Dex6 kDa surface. The thickness (d) and real refractive index (RI) of the lipid layer

3. RESULTS AND DISCUSSION 3.1. QCM Measurements. The QCM measurements in Figure 1 show that the liposomes of the three known lipid compositions spread as a bilayer on the SiO2 sensor surface. The three lipid compositions exhibited the typical adsorptionbursting behavior often seen in QCM experiments during vesicle spreading when SLBs are formed.9,31 The overlap of the normalized overtones (or more precisely, the lack of any difference between the overtones), the normalized frequency level of 25 Hz, and the fact that the changes in dissipation values are less than 2 × 10−6 (Supporting Information, Figure S4) indicates that good-quality lipid bilayers were formed.9 QCM measurements also revealed that the simple in situ wash

structures were calculated using the SPR Navi LayerSolver software v. 0.16 (BioNavis Ltd., Ylöjärvi, Finland). The software uses the wellknown Fresnel equation formalism for calculations28 but allows one to simultaneously process multiple SPR spectra in a single calculation, which has earlier been performed in several calculation steps.29,30 In this work optical modeling was performed using SPR angular spectra measured at 670 and 785 nm at the same time point and linking all thicknesses as common variables, while the complex refractive index was put either as an independent variable (for background) or as a linearly dependent variable between the two wavelengths used.30 2802

dx.doi.org/10.1021/la4046622 | Langmuir 2014, 30, 2799−2809

Langmuir

Article

Figure 3. QCM Δf/N vs time sensograms at F3, F5, F7, and F9 for all three vesicle formulations used in this study during interaction with the PEGSAM surface. (Left to right) Red, EggPC; black, EggPC+PS; blue, EggPC+PS+Chol). Horizontal line marks the position where Δf/N = 25 Hz. Arrows indicate events in the sensogram: (a) injection of liposomes, (b) end of injection of liposomes, (c) injection of water, and (d) end of injection of water. H2O treatment could not trigger bilayer formation, unlike with the Dex6 kDa surface. The surface clearly does not allow for bilayer formation, as the final frequency is far from the typical bilayer frequency (25 Hz) and the overtones do not overlap, which is a typical indication of a loose viscoelastic layer, such as a SVL.

demonstrated by Dodd et al.24 However, even though Dodd et al. managed to immobilize some kind of a supported lipid bilayer on SiO2 surface it was not a good-quality SLB according to their QCM-D results. However, it was possible to use a similar approach with small modifications in this study in order to spread the HepG2-extract mixed with EggPC into a bilayer. When interacting with the PEG-SAM surface, all vesicles in this study behaved completely differently than with the two previous surfaces. The PEG-SAM surface clearly promoted formation of SVLs, which can be seen from the significantly larger normalized overtone frequency changes compared to the typical 25 Hz for an SLB as well as from the fact that the normalized overtone frequencies do not overlap at all (Figure 3). This indicates that the deposited lipid layer was a viscoelastic layer composed of adsorbed vesicles. Furthermore, the H2O treatment that was capable of transforming the (partial) SVLs adsorbed on the Dex6 kDa surface did not result in formation of SLBs with the PEG-SAM surface. The dissipation results for the PEG-SAM surface also show a completely different behavior compared with the Dex6 kDa or SiO2 surfaces (Supporting Information, Section S3, Figure S6), further indicating that the PEG-SAM surface promotes SVL formation and not SLB formation. It should be noted that in order to properly prove SLB formation with the QCM, more than one measured overtone should be analyzed and compared, and the difference between the normalized overtones should be as small as possible. Apart from reaching a final frequency level of 25 Hz and negligible dissipation changes,31 the overlap of the normalized overtones is also an important indicator for showing the absence of liposomes, The sensitivity of a single overtone frequency for differentiating between an SLB and an SVL becomes smaller as the overtone frequency gets higher. At the 9th or 11th overtones the frequency changes for a 5 MHz crystal is nearly identical for an SLB and SVL due to the shear wave penetration depth of the higher overtone frequencies compared to the lower ones.26 For example, the shear wave penetration depth

cycle of CHAPS, Hellmanex II, ethanol, and ultrapure H2O used between consecutive measurements in this study was sufficient to clean the sensor completely, and it was not necessary to wash the sensors with piranha (conc. H2SO4:H2O2 3:1) as is commonly suggested.9 This enables a faster repetition of measurements and improves the safety of the preparative steps for vesicle spreading studies on SiO2 surfaces. A similar initial adsorption-bursting behavior seen on the SiO2 surface was also clearly seen for vesicles composed of known lipids, when they interact with the Dex6 kDa surface (Figure 2; red, black, and blue lines). On the other hand, the normalized frequency level and the overlap of the overtone frequencies clearly show that complete lipid bilayers were not spontaneously formed on the Dex6 kDa surface, and the deposited lipid layers remain as mixtures of SLBs and SLVs (Figure 2). However, treating the lipid layers deposited on the Dex6 kDa surface with H2O triggers a process which results in formation of almost perfect lipid bilayers, which is indicated by the normalized frequency levels of approximately 25 Hz and the overlap of the normalized overtone frequencies. The dissipation results also support these findings (Supporting Information, Section S3, Figure S5). The mechanism of the triggered SLB formation caused by H2O injection is not clear, and it seems to be specific for the lipid composition. It is probable that the mechanism is due to osmotic stress induced by the ionic strength gradient over the liposome membrane.32,33 H2O-triggered SLB formation could also be used to prepare an SLB from the vesicles composed of a mixture of the HepG2 cell membrane extract and EggPC (green lines in Figure 2). The higher level and poorer ovelap of the normalized overtone frequencies for the SLB of the HepG2-extract compared to the known lipids could be expected. This is because the HepG2 cell membrane extract contains intact cell membrane proteins, saccharides, and other components, which increase the size and viscoelasticity of the SLB of the HepG2-extract compared to the SLBs with known lipids. A similar approach of including cell membrane extract in an SLB (i.e., natural extract from Escheria coli bacteria mixed with fluid EggPC) has been previously 2803

dx.doi.org/10.1021/la4046622 | Langmuir 2014, 30, 2799−2809

Langmuir

Article

the SVL morphology both before and after water treatment according to the QCM modeling results. However, a decrease in thickness and an increase in elasticity indicate that some fraction of the liposomes on the PEG-SAM surface actually are fused into an SLB, but a significant portion of the adsorbed liposomes still retains the SVL morphology. It is also worth noting that the thicknesses of the SVLs on the Dex6 kDa surface before ultrapure water treatment are clearly smaller compared to the SVLs on the PEG-SAM surface. This is a strong indication that the liposomes deform to a larger extent on the Dex6 kDa surface than on the PEG-SAM surface, thus inducing larger stresses to the liposomes on the Dex6 kDa surface and therefore facilitating SLB formation when changing the osmotic conditions during ultrapure water treatment. One of the explanations for a larger deformation of liposomes on the Dex6 kDa surface compared to the PEG-SAM surface could be that there is a difference in the density and availability of the hydrophobic chains which causes differences in the interaction strength between the surfaces and the adsorbed liposomes. 3.1.1. Promotion of SLB or SVL Formation by the Substrate. The QCM results showed that there was a large difference between the Dex6 kDa and the PEG-SAM surfaces in promoting either SLB or SVL formation, even though both surfaces are relatively thin hydrogel supports with a thickness in the range of a few nanometers.The Dex6 kDa clearly promoted formation of SLBs, and the PEG-SAM promoted SVL formation. The most probable explanation for this is the type of surface morphology the polymers form on the sensor surface and the effect of different synthesis routes causing the lipidanchoring groups to be distributed differently in the Dex6 kDa and PEG-SAM layers. The measurement conditions used in this work were selected so that they would promote SLB formation on the SiO2 surface.9 The measurement protocols were also kept constant during all measurements, and it is unlikely that these would have a significant effect on the morphology control in this work. Furthermore, the Dex6 kDa and PEG-SAM surfaces should be noncharged due to the way they are produced, whereas the wettability of the surfaces shows a clear difference, i.e., contact angle of ultrapure water for the surfaces was 56° for PEG-SAM and 71° Dex6 kDa. Therefore, the surface charge cannot have a large role in promotion of either SLBs or SVLs on the different polymer surfaces synthesized in this study, while the wettability originating from the different distribution of hydrophobic linker chains in the polymer surfaces probably is more significant. The Dex6 kDa was formed from a highly concentrated solution, where the polymers are in a semientagled conformation and should also form a similar network upon surface linking.34 Furthermore, the hydrocarbon chains (i.e., decyl chains) were also postsynthesized into the formed dextran hydrogel, which most probably modifies mainly the outer surface portion of the hydrogel. Thus, the overall synthesis process should lead to a relatively dense and flat dextran surface with easily accessible hydrocarbon chains. The higher contact angle of Dex6 kDa compared to the PEG-SAM and the change in the contact angle as a result of the decyl binding to the Dex6 kDa also indicate that the decyl chains are more exposed in the Dex6 kDa than in the PEG-SAM surface. The PEG-SAM was formed by a self-assembly reaction with a custom-synthesized PEG polymer having premodified end groups. While the dominant species in the PEG polymer mixture were thiol and analogous disulfides of the PEGpolymer with one hydrocarbon chain (i.e., un-10-decene), the

for the third overtone for a 5 MHz crystal in pure water at 25 °C is 138 nm compared to 79 nm for the 9th overtone. QCM thickness analysis of the SLBs formed on the SiO2 and Dex6 kDa surfaces was performed using the Sauerbray equation. Additionally, the thickness of the SLB/SVL formed both before and after water treatment on both the Dex6 kDa and the PEG-SAM surfaces was analyzed using the viscoelastic model. The results of the analysis, i.e., thickness from Sauerbray analysis, and the thickness, elastic modulus (ε), and viscous modulus (η) from the viscoelastic analysis are presented in Table 1. Table 1. Result from the QCM Thickness Modelinga d (nm/ Sau) SiO2

Dex6 kDa Bef H2O

Dex6 kDa Aft H2O

PEG-SAM Bef H2O PEG-SAM Aft H2O

EggPC EggPC+PS EggPC+Chol EggPC EggPC+PS EggPC+Chol HepG2 EggPC EggPC+PS EggPC+Chol HepG2 EggPC EggPC+PS EggPC+Chol EggPC EggPC+PS EggPC+Chol

d (nm/VE)

ε (MPa)

η (Pa s)

7.5 8.3 10 16 5.7 5.8 4.5 6.1 14.5 17.0 16.0 11.8 11.8 9.8

0.85 0.56 0.55 0.72 1.75 1.20 1.55 0.66 0.73 0.69 0.63 0.79 0.71 0.80

0.0025 0.0014 0.00235 0.00365 0.0039 0.0040 0.0056 0.0028 0.00285 0.00265 0.00247 0.00260 0.00240 0.00245

4.7 4.5 4.7

4.9 5.3 4.4 2b

a

Sau = Sauerbray equation and V-E = viscoelastic modeling (equivalence circuit modeling). bThe Sauerbray equation produces false values because it assumes a rigid film, but as shown from the viscoelastic modeling, the HepG2 extract seems to behave as an SLB with protruding proteins.

The Sauerbrey QCM thickness analysis shows that approximately equal SLBs from all known lipid compositions were readily formed on the SiO2 surfaceas, whereas SLB formation on the Dex6 kDa surface only takes place after ultrapure water treatment. Applying the viscoelastic modeling to the lipid layers on the Dex6 kDa surface after water treatment produced systematically slightly larger thicknesses compared to the Sauerbrey analysis. The thickness and elastic moduli of the SLBs on the Dex6 kDa surface were generally significantly higher than those of the SVL layers on Dex6 kDa before water treatment, which supports the conclusion that SLBs are formed on the Dex6 kDa surface after water treatment. The thickness of the HepG2 extract mixture on the Dex6 kDa surface after ultrapure water treatment is slightly higher, while the elasticity is lower compared to the thickness of the known lipid mixtures. However, the thickness for the HepG2 extract mixture after ultrapure water treatment is still significantly smaller than any of the lipid mixtures before water treatment. The HepG2 results are reasonable when considering that there are membrane-protruding proteins present in the SLB formed from the HepG2 mixture which easily could contribute to a thickness increase of 1−2 nm and a higher viscoelasticity due to protruding proteins incorporated in the bilayer which traps water. The PEG-SAM surface clearly retains 2804

dx.doi.org/10.1021/la4046622 | Langmuir 2014, 30, 2799−2809

Langmuir

Article

Figure 4. SPR angle vs time sensograms measured with a wavelength of 670 nm during interaction of all five lipid vesicle formulations with the three different surfaces used in this study. (Left to right) SiO2 surface, PEG-SAM, Dex6 kDa, and HepG2 membrane extract on Dex6 kDa. Different lipid compositions are marked with colors as follows: EggPC, red ; EggPC+PS, black; EggPC+PS+Chol, blue; EggPC + Label, orange; HepG2-extact, green. Arrows indicate events in the sensogram: (a) injection of liposomes, (b) end of injection of liposomes, (c) injection of water, and (d) end of injection of water. SPR is sensitive to the refractive index of the lipids (density and packing), as well as to the effect of the supporting surface structure on the evanescent field (Supporting Information, Section S4, Table S2). This makes it relatively difficult to determine if a bilayer is formed or not. However, measurements with the label clearly show a large difference between the PEG-SAM and the other two surfaces, which correlates with the QCM measurements (Figures 1−3). The HepG2-extract has a higher SPR signal than the other lipid compositions on the Dex6 kDa surface because it also contains a higher amount of cholesterol, as well as membrane proteins and other cell membrane components that are not present in the known lipid compositions.

Section S4, Table S2).22 The difference between the signal level of SiO2 and Dex6 kDa SLBs can be explained by the effect caused by the support structure on the optical signal of the SLB. In an optical model an SLB adsorbed on a spacer hydrogel support structure gives approximately 15% less signal compared with an SLB adsorbed on a dense SiO2 surface (Supporting Information, Section S4, Table S2). This was approximately also the difference observed in the actual SPR measurements (Figure 4). SPR measurements with the vesicles containing the dye label did not largely differ from the pure EggPC lipid composition on the surfaces that were expected to promote SLB formation (i.e., SiO2 and Dex6 kDa−EggPC red trace and label+EggPC orange trace in Figure 4). However, the difference in the SPR signal between the EggPC and the label+EggPC lipid compositions during adsorption to the PEG-SAM surface was large. This is expected in the case of SVLs, because when a compound that absorbs light is trapped in the vesicles this will appear as a much higher RI index compared to vesicles without the lightabsorbing compound. This is due to the anomalous behavior of the refractive index when compounds that absorb light in the same wavelength region used for the SPR measurements are present.37−39This further confirms the earlier findings from QCM measurements, i.e., that the SiO2 and Dex6 kDa coatings promote SLB formation and PEG-SAM coating promotes SVL formation. The small difference between the SPR signal levels of the Dex6 kDa and SiO2 surface for the EggPC and label +EggPC lipid formulations is probably due to the extra space provided by the hydrogel under the SLB, which traps a small amount of the labeled material in the hydrogel. On the other hand, in the case of the SiO2 surface the SLB is formed on a rigid surface with a minimal space between the surface and the SLB, which does not allow any labeled material to be trapped between the surface and the SLB.

PEG polymer mixture also contained dithiols and unthiolated polymers. The concentration of the PEG polymer mixture was relatively high during the PEG-SAM formation but was still within a range where polymers behave mostly as individual coils rather than as an entangled network.34 Due to steric interactions the self-assembly of polymers under such conditions should form a more “mushroom”-like surface with individual polymer coils rather than a dense polymer brush structure.35 This kind of behavior is also supported by another study where similar PEG self-assembly was studied with SPR for PEG thickness and density as a function of molecular weight, which is also proportional to polymer coil dimensions.36 Taken together, promotion of SLB formation on the Dex6 kDa and SVL formation on the PEG-SAM surface is quite reasonable. The brushlike Dex6 kDa hydrogel most probably forms a rather homogeneous and smooth surface with a sufficient amount of easily accessible hydrocarbon chains, which promotes SLB formation, whereas the mushroom-like PEGSAM surface is probably not so homogeneous and smooth with less accessible hydrocarbon chains, thus promoting the vesicles to retain their shape on the surface. 3.2. SPR Measurements. SPR measurements did not show as pronounced bursting behavior for any of the surface/lipid compositions as was seen in the QCM measurements (Figure 4). This is reasonable as the QCM technique is sensitive to the water content in the adsorbed layer while SPR is not, and SVLs carry a significant amount of water trapped in the liposomes. In spite of this, a small bursting behavior which follows the general shape of the SLB formation steps in QCM measurements was detectable in the SPR measurements. The level of the SPR signal acquired during deposition of the vesicles corresponds well with the expected values for a lipid bilayer when evaluated with an optical model of the systems (Supporting Information, 2805

dx.doi.org/10.1021/la4046622 | Langmuir 2014, 30, 2799−2809

Langmuir

Article

The label+EggPC SPR measurements also show that it was not necessary to use the water-triggering step used in the QCM measurements for SPR measurements in order to form SLBs of the known lipid compositions on the Dex6 kDa surface. This is probably due to the fact that the shear stress in the SPR flow channel is actually higher compared to the QCM flow channel, because the heights of the two flow channels used in the calculations for synchronizing the flow conditions are in reality not as well defined and precise as assumed. On the other hand, the HepG2-extract lipid composition still required a watertriggering step in the SPR measurements in order to form an SLB on the Dex6 kDa surface. Furthermore, there was a clear difference in the absolute SPR signal level measured for the SLB of the HepG2-extract lipid composition (green line in Figure 4) compared to the other lipid compositions. The reason for this is that the HepG2-extract lipid composition contains more cholesterol as well as membrane proteins and other cell membrane components which are not present in the known lipid compositions. This also makes the vesicles prepared from the HepG2-extract lipid composition more stable and viscoelastic,40 which consequently requires some triggering step such as the water treatment to induce SLB formation. 3.2.1. Two-Wavelength SPR Analysis. A two-wavelength analysis for determining the thickness (d) and refractive index (RI) of the adsorbed lipid structures was performed with the LayerSolver software for all known lipid compositions. Background parameters were calculated from the spectra taken 1 min before vesicle injection, and sample spectra were taken 10 min after vesicle injection was ended. The optical parameters obtained from the background spectra and examples of the optical fits to the background spectra can be found in the Supporting Information, Table S1 and Figure S7. It was not possible to accurately determine or separate the optical properties of the Dex6 kDa and PEG-SAM layers in the background spectra, because their contribution to the overall background spectra was very small. Hence, the optical parameters for the background spectra in the case of Dex6kD and PEG-SAM surfaces should be considered as “apparent” parameters, which means that they are not physically correct but are still accurate enough for analysis purposes in this study. This in combination with the effect caused by the support structure on the optical signal discussed above (Supporting Information, Section S4, Table S2) means that the results from the two-wavelength SPR analysis for the lipid layer structure formed on the Dex6 kDa and PEG-SAM surfaces should only be considered as qualitative and reflect only the differences inside the same series. The results from the two-wavelength SPR analysis for the lipid layer structures formed on the SiO2 surface do not suffer from this drawback and can be viewed as accurate. The results of the two-wavelength SPR analysis for the different lipid compositions and support surfaces are presented in Table 2. The d and RI values obtained for the SLBs on SiO2 in this study correlate well with results obtained with a similar surface in dual-polarization interferometry by Lee et al.,22 especially when taking into account the difference in the lipids used. The lipid mixtures used in this study were natural extracts (EggPC) which are mixtures of many components with a high amount of unsaturated fatty acid chains, while the work of Lee et al. was performed with fully saturated dimyristoyl phosphatidylcholine (DMPC) as the main component. It should also be noted that while the reference in the work by Lee et al. also relies on

Table 2. Lipid Layer Parameters Obtained from Optical Modeling of SPR Measurements surface SiO2 SiO2 SiO2 Dex6 kDa Dex6 kDa Dex6 kDa PEGSAM PEGSAM PEGSAM PEGSAM

lipid

d

n (670 nm)

n (785 nm)

dn/dL (−1/ nm)

4.70 4.63 5.61

1.4421 1.4431 1.4236

1.4381 1.4387 1.4191

0.000035 0.000038 0.000039

EggPC EggPC+PS EggPC +Chol EggPC EggPC+PS EggPC +Chol EggPC

5.11 5.16 6.38

1.4735 1.4729 1.4401

1.4680 1.4673 1.4371

0.000048 0.000049 0.000026

10.79

1.3977

1.3975

0.00002

EggPC+PS

13.84

1.3813

1.3804

0.000008

8.69

1.3928

1.3882

0.000040

10.55

1.4031

1.4013

0.000016

EggPC +Chol EggPC +Label

results obtained from a secondary technique (i.e., neutron scattering for thickness determination) for analyzing both the refractive index and the thickness for their lipid mixtures, we obtained similar results in this study without a need for similar references. Both the current study and the literature show that the cholesterol-containing lipid mixture has a higher thickness, which can be explained by the condensing effect of cholesterol in the SLB. Cholesterol situates between the fatty acid tails of the phospholipids in the SLB and forces them to stand up straighter than they would without the presence of cholesterol.41 This effect was more pronounced with the longer unsaturated chains in this work compared with saturated alkane chains used by Lee et al. As discussed above, the d and RI values obtained for the actual Dex6 kDa and PEG-SAM surfaces have to be taken as qualitative rather than quantitative. Even so, the d and RI values obtained for the lipid layers on the Dex6 kDa surface support the fact that an SLB was formed with all different lipid compositions used in this study, which is further supported by the QCM results. The d and RI values for the SLBs on the Dex6 kDa surface were slightly higher but within a reasonable range when compared to the values obtained for the SiO2 surface. This might depend on the influence of the background during optical modeling as discussed above, or alternatively the SLBs might ripple on the Dex6 kDa surface, consequently displaying a slightly larger thickness compared to the SiO2 surface. It might also be possible that the higher RI obtained for the SLBs on the Dex6 kDa surface compared to the SiO2 surface indicates that the lipids can pack more tightly on the Dex6 kDa surface than on the SiO2. However, this can only be speculated, as the results obtained from optical modeling were too uncertain for drawing stronger conclusions, because of the issues related to the background uncertainty and the spacing layer sensitivity (Supporting Information, section S4). The lipid layers on the PEG-SAM surface clearly shows higher d and lower RI compared to the two other surfaces, which was expected on the basis of the QCM results. This supports the earlier findings that the PEG-SAM surface rather promotes SVL than SLB formation. However, the d’s were quite small compared to the vesicle sizes. This can be explained either by the uncertainty in the background used during optical modeling or most probably because the SVLs deform into disclike vesicles upon absorption on the PEG-SAM surface which 2806

dx.doi.org/10.1021/la4046622 | Langmuir 2014, 30, 2799−2809

Langmuir

Article

Figure 5. Normalized SPR minimum angle vs time sensorgrams of air stability measurements of the EggPC+PS+Chol lipid mixture adsorbed on the Dex6kDA surface. Three repetition sensorgrams have been normalized to the deposition plateau (interval starting at 20−25 min), and the air stability in percentage has been characterized from the magnitude of the returning signal (interval at 45−50 min). Air treatment was also recorded online but was cut out of the sensogram for clarity. The presence of air in the flow channel was confirmed by the shift in the TIR angle during the measurement (not shown). Three repetitions gave an average of 86% stability of the deposited EggPC+PS+Chol SLB.

consequently is reflected as a much smaller thickness than would be expected for full nondeformed vesicles. The results from two-wavelength SPR analysis correlate very well with the results obtained from QCM thickness analysis. Hence, these SPR results support the conclusions drawn from the QCM results on SLB and SVL formation on the different surfaces used in this study. The results are also supported well by earlier studies with similar experiments reported in the literature.22 While the QCM was an easier method for detecting the SLB formation as such, it is quite apparent that the twowavelength SPR analysis approach was more sensitive to slight changes in the lipid layer thickness and density than the QCM experiments even though the SPR analysis was performed in a qualitative intraseries manner. However, together these two methods seem to offer an excellent combination for studying the biophysical properties of different membranes and other adsorbed or layered materials. 3.2.2. Air Stability of the SLB. The air stability for the EggPC-PS-Chol lipid mixture on the Dex6 kDa surface was tested in triplicate by flowing air through the SPR microfluidic system at a nominal speed of 200 μL/min. The SPR flow channels have an internal volume of 1 μL, meaning that the gas content in the flow channel was replenished 3.33 times/s. Other lipid mixtures or surfaces used in this study were not tested, because SLBs formed on the SiO2 surface are known to be unstable during transition through the air−water interface.16 The air stability of the SVLs adsorbed on the PEG-SAM surface was not studied because SLBs are often more desirable biomimetic surfaces than SVLs for use with surface-sensitive detection techniques because of their well-defined structures, which consequently makes analysis of the signal responses with SLBs much easier compared to SVLs. The less complex lipid mixtures were also left out from the air stability studies because the main challenge in preparing biomimetic surfaces by vesicle adsorption is to form air-stable SLBs with cholesterolcontaining vesicles as well as for vesicles with lipid

compositions resembling natural cell membranes as close as possible. When the SPR flow channel was filled with air this caused a large transition in the SPR angular spectra, where the SPR peak typical for water was transformed into a multimodal mixed angular spectrum. During this time it was not possible to track the SPR peak minimum. However, the crossing of the air− water interface was clearly indicated by the shift of the critical angle of the SPR peak to the range typical for water. Due to the difficulty in tracking the SPR peak during air exposure, the time period for the air flow in the measurements was cut from the SPR sensograms (Figure 5). Figure 5 clearly shows that the SPR signal levels returned close to the original signal level after air treatment of the EggPC+PS+Chol SLB. An air stability of 86% was obtained for the EggPC+PS+Chol SLB by comparing the averaged SPR signal level before air injection to the averaged SPR signal level after air treatment. The measurements were performed with a constant flow of air through the flow channel for 10 min, which means that the EggPC+PS +Chol SLB was also continuously drying during the experiment. This indicates that the EggPC+PS+Chol SLB on the Dex6 kDa surface had an excellent resistance against drying upon exposure to air and upon thransition through the air− water interface. 3.3. Comparison of Surfaces. The QCM and SPR results indicate that the Dex6 kDa surface offers a platform for SLB deposition similar to the SiO2 surface, which was utilized as a reference surface in this study. The PEG-SAM surface however promotes SVL deposition. While both the Dex6 kDa and the PEG-SAM surfaces are hydrophilic and highly hydrated, the difference in promoting different lipid morphology was most probably due to the differences in the inherent structure, density, and location of the hydrophobic groups of the hydrogels. Both of the hydrogel-based sensor surface coatings could be regenerated by a simple CHAPS detergent washing cycle, while 2807

dx.doi.org/10.1021/la4046622 | Langmuir 2014, 30, 2799−2809

Langmuir

Article

with the well-established QCM technique as well as with other optical techniques used for SLB characterization. This study also shows that the combination of both SPR and QCM methodologies provides high-quality data with a large amount of information for characterizing nanometer-scale systems. It is apparent that use of both techniques is especially beneficial in biophysics and similar applications, due to the quality and magnitude of the information required for these applications.

the SiO2 surface needed a much harsher and complex washing procedure before reuse. This is also a beneficial property for screening of interactions, as it allows a faster experimental cycle and increases the throughput of the assay. The throughput, along with a simpler automation caused by the less-complicated wash cycle, makes the new Dex6 kDa and PEG-SAM surfaces viable candidates for wide applications in the screening of biochemical interactions and in pharmaceutical research. This kind of a reusable sensor structure offers even greater benefits in several application areas when compared to nonregenerable sensor structures, e.g., biotin−avidin-based sensors. The air stability exhibited by the SLBs on the Dex6 kDa surface was found to be good. This property is important in a wider application of SLBs within interaction research and pharmaceutical development, as it allows resistance to sampling mistakes (i.e., dissolved gases outgassing in microfluidics), more advanced samples (air-bubble-trapped samples), or even a predeposition of SLBs and storage of them. A similar air stability has also been shown for supports containing a Langmuir−Blodgett deposited monolayer with a Zr2+-ion top layer, which actually required that palmitoyl-oleyl phosphatic acid is included into the SLB mixture.18 The benefits of the SLB supported with Dex6 kDa over the Zr2+-ion approach are a wider lipid functionality, a regenerability, and a simpler synthesis. Another similar approach has been introduced using cholesterols in a PEG matrix as linking groups.17 This approach was similar to the Dex6 kDa approach in this study, but the air stability was performed in much milder conditions (i.e., careful transfer of the substrate and SLB through air− water interface), while in this study the stability was demonstrated under a constant air flow. An approach where SLBs have been “sandwiched” between proteins or PEG has shown good results for the SLB air stability. However, the approach often requires certain groups or macromolecules to be added to the SLB formulation.16,19 This can cause interference in measuring the interactions or other properties of the SLBs, which further makes the Dex6 kDa surface developed in this study a more viable approach for preparing SLBs for any applications studying these properties.

■

ASSOCIATED CONTENT

S Supporting Information *

Polymer end group modification of PEG; DLS size distribution graph; impedance-QCM dissipation results; SPR signal dependency on coating material; optical constants of two-wavelength SPR analysis. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: niko.granqvist@helsinki.fi. *E-mail: tapani.viitala@helsinki.fi. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research was funded by Academy of Finland (Project Nos. 137053, 140980, and 263861). N.G. thanks Foundation of Anja and Mirja Tissari (2012) and Foundation of Elli Turunen (2013), both member foundations of the Finnish Cultural Foundation (Suomen Kulttuurirahasto), for support. We also thank M.Sc. Jussi Tuppurainen from BioNavis Ltd. for his assistance in the two-wavelength SPR fitting and Dr. Mohammed Elmowafy for extracting the HepG2 cell membranes.

■

REFERENCES

(1) Krishna;, R.; Yu, L. Biopharmaceutic Applications in Drug Development; Springer: New York, 2008. (2) Avdeef, A. Absorption and Drug Development; John Wiley & Sons: New York, 2003. (3) Flaten, G. E.; Dhanikula, A. B.; Luthman, K.; Brandl, M. Drug permeability across a phospholipid vesicle based barrier: A novel approach for studying passive diffusion. Eur. J. Pharm. Sci. 2006, 27 (1), 80−90. (4) Hayeshi, R.; Hilgendorf, C.; Artursson, P.; Augustijns, P.; Brodin, B.; Dehertogh, P.; Fisher, K.; Fossati, L.; Hovenkamp, E.; Korjamo, T.; Masungi, C.; Maubon, N.; Mols, R.; Mullertz, A.; Monkkonen, J.; O’Driscoll, C.; Oppers-Tiemissen, H. M.; Ragnarsson, E. G.; Rooseboom, M.; Ungell, A. L. Comparison of drug transporter gene expression and functionality in Caco-2 cells from 10 different laboratories. Eur. J. Pharm. Sci. 2008, 35 (5), 383−96. (5) Volpe, D. A. Variability in Caco-2 and MDCK cell-based intestinal permeability assays. J. Pharm. Sci. 2008, 97 (2), 712−25. (6) Scannell, J.; Blanckley, A.; Boldon, H.; Warrington, B. Diagnosing the decline in pharmaceutical R&D efficiency. Nat. Rev. Drug Discovery 2012, 11 (3), 191−200. (7) Cooper, M. Advances in membrane receptor screening and analysis. J. Mol. Recognit. 2004, 17 (4), 286−315. (8) Castellana, E. T.; Cremer, P. S. Solid supported lipid bilayers: From biophysical studies to sensor design. Surf. Sci. Rep. 2006, 61 (10), 429−444. (9) Cho, N. J.; Frank, C. W.; Kasemo, B.; Hook, F. Quartz crystal microbalance with dissipation monitoring of supported lipid bilayers on various substrates. Nat. Protoc. 2010, 5 (6), 1096−106.

4. CONCLUSIONS This work describes the synthesis of two new surface coatings that promotes different lipid layer morphologies during vesicle spreading on the supporting surface. The low-molecular-weight dextran-based Dex6 kDa surface showed good performance in promoting SLB formation with both known lipids as well as with a mixture containing the natural membrane extract of HepG2 cells. The low-molecular-weight self-assembled PEGSAM surface promoted SVL formation on the surface with all lipid compositions tested in this study. It was also possible to demonstrate that the Dex6 kDa surface effectively protects the SLB layer from dehydration after deposition. The air stability allows one to develop more robust SLB-based experimental assays, thereby increasing the probability to further develop SLBs as useful platforms for a wide range of applications. The simple synthesis procedures of a surface combined with air stability for the deposited lipid layers makes it possible to achieve more flexible and robust SLBs for applications in areas such as drug discovery and development, biosensing, and biophysics. In addition, this study further demonstrates utilization of the relatively rare multiple-wavelength SPR method. The twowavelength SPR analysis used provided results which correlated 2808

dx.doi.org/10.1021/la4046622 | Langmuir 2014, 30, 2799−2809

Langmuir

Article

a refractometer for liquids and ultrathin films. Sens. Actuators, B 2010, 149 (1), 212−220. (30) Granqvist, N.; Liang, H.; Laurila, T.; Sadowski, J.; Yliperttula, M.; Viitala, T. Characterizing ultrathin and thick organic layers by surface plasmon resonance three-wavelength and waveguide mode analysis. Langmuir 2013, 29 (27), 8561−8571. (31) Keller, C.; Glasmästar, K.; Zhdanov, V.; Kasemo, B. Formation of supported membranes from vesicles. Phys. Rev. Lett. 2000, 84 (23), 5443−5446. (32) Erik, R.; Fredrik, H.; Bengt, K., Intact Vesicle Adsorption and Supported Biomembrane Formation from Vesicles in Solution: Influence of Surface Chemistry, Vesicle Size, Temperature, and Osmotic Pressure †. Langmuir 2003, 19. (33) Hain, N.; Gallego, M.; Reviakine, I. Unraveling supported lipid bilayer formation kinetics: osmotic effects. Langmuir 2013, 29 (7), 2282−2288. (34) Burchard, W. Solution properties of branched macromolecules. In Branched Polymers II; Roovers, J., Ed.; Springer: Berlin, 1999; Vol. 143, pp 113−194. (35) Balazs, A. C.; Singh, C.; Zhulina, E.; Chern, S.-S.; Lyatskaya, Y.; Pickett, G. Theory of polymer chains tethered at interfaces. Prog. Surf. Sci. 1997, 55 (3), 181−269. (36) Schoch, R. L.; Lim, R. Y. H. Non-Interacting Molecules as Innate Structural Probes in Surface Plasmon Resonance. Langmuir 2013, 29 (12), 4068−4076. (37) Esteban, Ó .; González-Cano, A.; Díaz-Herrera, N.; Navarrete, M.-C. Absorption as a selective mechanism in surface plasmon resonance fiber optic sensors. Opt. Lett. 2006, 31 (21), 3089−3091. (38) Hanning, A.; Roeraade, J.; Delrow, J. J.; Jorgenson, R. C. Enhanced sensitivity of wavelength modulated surface plasmon resonance devices using dispersion from a dye solution. Sens. Actuators, B 1999, 54 (1−2), 25−36. (39) Nakkach, M.; Lecaruyer, P.; Bardin, F.; Sakly, J.; Lakhdar, Z. B.; Canva, M. Absorption and related optical dispersion effects on the spectral response of a surface plasmon resonance sensor. Appl. Opt. 2008, 47 (33), 6177−6182. (40) Fidorra, M.; Duelund, L.; Leidy, C.; Simonsen, A.; Bagatolli, L. Absence of fluid-ordered/fluid-disordered phase coexistence in ceramide/POPC mixtures containing cholesterol. Biophys. J. 2006, 90 (12), 4437−4451. (41) Ipsen, J. H.; Mouritsen, O. G.; Bloom, M. Relationships between lipid membrane area, hydrophobic thickness, and acyl-chain orientational order. The effects of cholesterol. Biophys. J. 1990, 57 (3), 405− 412.

(10) Rich, R.; Myszka, D. Grading the commercial optical biosensor literature-Class of 2008: ‘The Mighty Binders’. J. Mol. Recognit. 2010, 23 (1), 1−64. (11) Baird, C.; Courtenay, E.; Myszka, D. Surface plasmon resonance characterization of drug/liposome interactions. Anal. Biochem. 2002, 310 (1), 93−99. (12) Deniaud, A.; Rossi, C.; Berquand, A.; Homand, J.; Campagna, S.; Knoll, W.; Brenner, C.; Chopineau, J. Voltage-dependent anion channel transports calcium ions through biomimetic membranes. Langmuir 2007, 23 (7), 3898−3905. (13) Keller, C.; Kasemo, B. Surface specific kinetics of lipid vesicle adsorption measured with a quartz crystal microbalance. Biophys. J. 1998, 75 (3), 1397−1402. (14) Rossi, C.; Chopineau, J. Biomimetic tethered lipid membranes designed for membrane-protein interaction studies. Eur. Biophys. J. 2007, 36 (8), 955−65. (15) Branden, M.; Dahlin, S.; Hook, F. Label-free measurements of molecular transport across liposome membranes using evanescentwave sensing. ChemPhysChem 2008, 9 (17), 2480−5. (16) Holden, M.; Jung, S.-Y.; Yang, T.; Castellana, E.; Cremer, P. Creating fluid and air-stable solid supported lipid bilayers. J. Am. Chem. Soc. 2004, 126 (21), 6512−6513. (17) Deng, Y.; Wang, Y.; Holtz, B.; Li, J.; Traaseth, N.; Veglia, G.; Stottrup, B.; Elde, R.; Pei, D.; Guo, A.; Zhu, X. Y. Fluidic and air-stable supported lipid bilayer and cell-mimicking microarrays. J. Am. Chem. Soc. 2008, 130 (19), 6267−6271. (18) Fabre, R. M.; Talham, D. R. Stable supported lipid bilayers on zirconium phosphonate surfaces. Langmuir 2009, 25 (21), 12644−52. (19) Albertorio, F.; Diaz, A.; Yang, T.; Chapa, V.; Kataoka, S.; Castellana, E.; Cremer, P. Fluid and air-stable lipopolymer membranes for biosensor applications. Langmuir 2005, 21 (16), 7476−7482. (20) Summanen, M.; Granqvist, N.; Tuominen, R.; Yliperttula, M.; Verrips, C.; Boonstra, J.; Blanchetot, C.; Ekokoski, E., Kinetics of PKCε activating and inhibiting llama single chain antibodies and their effect on PKCε translocation in HeLa cells. PLoS ONE 2012, 7, (4). (21) Löfås, S.; Johnsson, B. A novel hydrogel matrix on gold surfaces in surface plasmon resonance sensors for fast and efficient covalent immobilization of ligands. J. Chem. Soc., Chem. Commun. 1990, No. 21, 1526−1528. (22) Lee, T.-H.; Heng, C.; Swann, M.; Gehman, J.; Separovic, F.; Aguilar, M.-I. Real-time quantitative analysis of lipid disordering by aurein 1.2 during membrane adsorption, destabilisation and lysis. Biochim. Biophys. Acta 2010, 1798 (10), 1977−1986. (23) Céline, E.-C.; Ophélie, F.; Jacques, P.; Jean-Claude, M.; Christian, B., Self-Assembly of Solid-Supported Membranes Using a Triggered Fusion of Phospholipid-Enriched Proteoliposomes Prepared from the Inner Mitochondrial Membrane 1. Langmuir 2005, 21. (24) Dodd, C.; Johnson, B.; Jeuken, L.; Bugg, T.; Bushby, R.; Evans, S., Native E. coli inner membrane incorporation in solid-supported lipid bilayer membranes. Biointerphases 2008, 3, (2). (25) Lee, T.-H.; Hall, K.; Swann, M.; Popplewell, J.; Unabia, S.; Park, Y.; Hahm, K.-S.; Aguilar, M.-I. The membrane insertion of helical antimicrobial peptides from the N-terminus of Helicobacter pylori ribosomal protein L1. Biochim. Biophys. Acta 2010, 1798 (3), 544− 557. (26) Bandey, H. L.; Martin, S. J.; Cernosek, R. W.; Hillman, A. R. Modeling the Responses of Thickness-Shear Mode Resonators under Various Loading Conditions. Anal. Chem. 1999, 71 (11), 2205−2214. (27) Viitala, T.; Liang, H.; Gupta, M.; Zwinger, T.; Yliperttula, M.; Bunker, A. Fluid dynamics modeling for synchronizing surface plasmon resonance and quartz crystal microbalance as tools for biomolecular and targeted drug delivery studies. J. Colloid Interface Sci. 2012, 378 (1), 251−9. (28) Albers, W.; Vikholm-Lundin, I. Surface Plasmon Resonance on Nanoscale Organic Films. In Nano-Bio-Sensing; Carrara, S., Ed.; Springer: New York, 2010. (29) Liang, H.; Miranto, H.; Granqvist, N.; Sadowski, J. W.; Viitala, T.; Wang, B.; Yliperttula, M. Surface plasmon resonance instrument as 2809

dx.doi.org/10.1021/la4046622 | Langmuir 2014, 30, 2799−2809